Disease Gene Prioritization Based on Topological Similarity in Protein-Protein Interaction Networks

In recent years, many algorithms have been developed to narrow down the set of candidate disease genes implicated by genome wide association studies (GWAS), using knowledge on protein-protein interactions (PPIs). All of these algorithms are based on a common principle; functional association between proteins is correlated with their connectivity/proximity in the PPI network. However, recent research also reveals that networks are organized into recurrent network schemes that underlie the mechanisms of cooperation among proteins with different function, as well as the crosstalk between different cellular processes. In this paper, we hypothesize that proteins that are associated with similar diseases may exhibit patterns of “topological similarity” in PPI networks. Motivated by these observations, we introduce the notion of “topological profile”, which represents the location of a protein in the network with respect to other proteins. Based on this notion, we develop a novel measure to assess the topological similarity of proteins in a PPI network. We then use this measure to develop algorithms that prioritize candidate disease genes based on the topological similarity of their products and the products of known disease genes. Systematic experimental studies using an integrated human PPI network and the Online Mendelian Inheritance (OMIM) database show that the proposed algorithm, VAVIEN, clearly outperforms state-of-the-art network based prioritization algorithms. VAVIEN is available as a web service at http://www.diseasegenes.org.